feat(imageutil): add upscale and numbered grid-overlay preprocessing

2026-10-12 00:45:28 +00:00 · 2026-06-07 17:32:35 +01:00
parent bfbb78383d
commit a49c6ab08d
4 changed files with 260 additions and 1 deletions
@@ -2,7 +2,10 @@ module github.com/lukaszraczylo/captcha-solver-mcp

 go 1.26.3

-require github.com/modelcontextprotocol/go-sdk v1.2.0
+require (
+	github.com/modelcontextprotocol/go-sdk v1.2.0
+	golang.org/x/image v0.41.0
+)

 require (
 	github.com/google/jsonschema-go v0.3.0 // indirect
@@ -8,6 +8,8 @@ github.com/modelcontextprotocol/go-sdk v1.2.0 h1:Y23co09300CEk8iZ/tMxIX1dVmKZkzo
 github.com/modelcontextprotocol/go-sdk v1.2.0/go.mod h1:6fM3LCm3yV7pAs8isnKLn07oKtB0MP9LHd3DfAcKw10=
 github.com/yosida95/uritemplate/v3 v3.0.2 h1:Ed3Oyj9yrmi9087+NczuL5BwkIc4wvTb5zIM+UJPGz4=
 github.com/yosida95/uritemplate/v3 v3.0.2/go.mod h1:ILOh0sOhIJR3+L/8afwt/kE++YT040gmv5BQTMR2HP4=
+golang.org/x/image v0.41.0 h1:8wS72eGJMJaBxK6okTzd4WaXumUlTVlb753MlsSvTCo=
+golang.org/x/image v0.41.0/go.mod h1:uIc348UZMSvS5Z65CVZ7iDPaNobNFEPeJ4kbqTOszmA=
 golang.org/x/oauth2 v0.30.0 h1:dnDm7JmhM45NNpd8FDDeLhK6FwqbOf4MLCM9zb1BOHI=
 golang.org/x/oauth2 v0.30.0/go.mod h1:B++QgG3ZKulg6sRPGD/mqlHQs5rB3Ml9erfeDY7xKlU=
 golang.org/x/tools v0.34.0 h1:qIpSLOxeCYGg9TrcJokLBG4KFA6d795g0xkBkiESGlo=
@@ -1,12 +1,34 @@
 package imageutil

 import (
+	"bytes"
 	"encoding/base64"
+	"image"
+	_ "image/png" // register PNG decoder for test helpers
+	"image/color"
+	"image/png"
 	"os"
 	"path/filepath"
 	"testing"
 )

+// makePNG creates an in-memory PNG of the given size filled with a solid color.
+func makePNG(t *testing.T, w, h int) []byte {
+	t.Helper()
+	img := image.NewRGBA(image.Rect(0, 0, w, h))
+	c := color.RGBA{R: 100, G: 149, B: 237, A: 255}
+	for y := 0; y < h; y++ {
+		for x := 0; x < w; x++ {
+			img.SetRGBA(x, y, c)
+		}
+	}
+	var buf bytes.Buffer
+	if err := png.Encode(&buf, img); err != nil {
+		t.Fatalf("makePNG encode: %v", err)
+	}
+	return buf.Bytes()
+}
+
 func TestDecode(t *testing.T) {
 	raw := []byte{0x89, 0x50, 0x4e, 0x47} // PNG magic prefix
 	b64 := base64.StdEncoding.EncodeToString(raw)
@@ -36,6 +58,100 @@ func TestDecode(t *testing.T) {
 	})
 }

+func TestUpscale(t *testing.T) {
+	const srcW, srcH = 30, 30
+
+	t.Run("factor 3 doubles dims", func(t *testing.T) {
+		src := makePNG(t, srcW, srcH)
+		out, mt, err := Upscale(src, 3)
+		if err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+		if mt != "image/png" {
+			t.Fatalf("want image/png, got %q", mt)
+		}
+		img, _, err := image.Decode(bytes.NewReader(out))
+		if err != nil {
+			t.Fatalf("result not decodable: %v", err)
+		}
+		b := img.Bounds()
+		if b.Dx() != srcW*3 || b.Dy() != srcH*3 {
+			t.Fatalf("want %dx%d, got %dx%d", srcW*3, srcH*3, b.Dx(), b.Dy())
+		}
+	})
+
+	t.Run("factor 10 clamps to 6", func(t *testing.T) {
+		src := makePNG(t, srcW, srcH)
+		out, _, err := Upscale(src, 10)
+		if err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+		img, _, err := image.Decode(bytes.NewReader(out))
+		if err != nil {
+			t.Fatalf("result not decodable: %v", err)
+		}
+		b := img.Bounds()
+		if b.Dx() != srcW*6 || b.Dy() != srcH*6 {
+			t.Fatalf("want %dx%d (clamped to 6), got %dx%d", srcW*6, srcH*6, b.Dx(), b.Dy())
+		}
+	})
+
+	t.Run("factor 1 unchanged dims", func(t *testing.T) {
+		src := makePNG(t, srcW, srcH)
+		out, mt, err := Upscale(src, 1)
+		if err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+		if mt != "image/png" {
+			t.Fatalf("want image/png, got %q", mt)
+		}
+		img, _, err := image.Decode(bytes.NewReader(out))
+		if err != nil {
+			t.Fatalf("result not decodable: %v", err)
+		}
+		b := img.Bounds()
+		if b.Dx() != srcW || b.Dy() != srcH {
+			t.Fatalf("want %dx%d, got %dx%d", srcW, srcH, b.Dx(), b.Dy())
+		}
+	})
+}
+
+func TestAnnotateGrid(t *testing.T) {
+	const size = 90
+
+	t.Run("3x3 grid same dims", func(t *testing.T) {
+		src := makePNG(t, size, size)
+		out, err := AnnotateGrid(src, 3, 3)
+		if err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+		img, _, err := image.Decode(bytes.NewReader(out))
+		if err != nil {
+			t.Fatalf("result not decodable: %v", err)
+		}
+		b := img.Bounds()
+		if b.Dx() != size || b.Dy() != size {
+			t.Fatalf("want %dx%d, got %dx%d", size, size, b.Dx(), b.Dy())
+		}
+	})
+
+	t.Run("rows=0 returns valid image", func(t *testing.T) {
+		src := makePNG(t, size, size)
+		out, err := AnnotateGrid(src, 0, 3)
+		if err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+		img, _, err := image.Decode(bytes.NewReader(out))
+		if err != nil {
+			t.Fatalf("result not decodable: %v", err)
+		}
+		b := img.Bounds()
+		if b.Dx() != size || b.Dy() != size {
+			t.Fatalf("want %dx%d, got %dx%d", size, size, b.Dx(), b.Dy())
+		}
+	})
+}
+
 func TestGridCentroids(t *testing.T) {
 	pts := GridCentroids(300, 300, 3, 3)
 	if len(pts) != 9 {
@@ -0,0 +1,138 @@
+package imageutil
+
+import (
+	"bytes"
+	"fmt"
+	"image"
+	"image/color"
+	"image/draw"
+	_ "image/jpeg" // register JPEG decoder
+	"image/png"
+
+	xdraw "golang.org/x/image/draw"
+	"golang.org/x/image/font"
+	"golang.org/x/image/font/basicfont"
+	"golang.org/x/image/math/fixed"
+)
+
+// Upscale decodes a PNG/JPEG image, scales it by factor (clamped to 1..6)
+// using high-quality resampling, and returns a PNG re-encode plus "image/png".
+func Upscale(data []byte, factor int) ([]byte, string, error) {
+	if factor < 1 {
+		factor = 1
+	}
+	if factor > 6 {
+		factor = 6
+	}
+
+	src, _, err := image.Decode(bytes.NewReader(data))
+	if err != nil {
+		return nil, "", fmt.Errorf("upscale: decode: %w", err)
+	}
+
+	bounds := src.Bounds()
+	newW := bounds.Dx() * factor
+	newH := bounds.Dy() * factor
+	dst := image.NewRGBA(image.Rect(0, 0, newW, newH))
+
+	xdraw.CatmullRom.Scale(dst, dst.Bounds(), src, src.Bounds(), xdraw.Over, nil)
+
+	var buf bytes.Buffer
+	if err := png.Encode(&buf, dst); err != nil {
+		return nil, "", fmt.Errorf("upscale: encode: %w", err)
+	}
+	return buf.Bytes(), "image/png", nil
+}
+
+// AnnotateGrid decodes the image and overlays a rows x cols grid: thin lines
+// on cell boundaries and the 0-based row-major cell index drawn in the
+// top-left of each cell, returning a PNG. Helps a vision model reference cells
+// unambiguously. If rows<=0 or cols<=0, returns the original re-encoded PNG.
+func AnnotateGrid(data []byte, rows, cols int) ([]byte, error) {
+	src, _, err := image.Decode(bytes.NewReader(data))
+	if err != nil {
+		return nil, fmt.Errorf("annotategrid: decode: %w", err)
+	}
+
+	bounds := src.Bounds()
+	dst := image.NewRGBA(bounds)
+	draw.Draw(dst, bounds, src, bounds.Min, draw.Src)
+
+	if rows <= 0 || cols <= 0 {
+		var buf bytes.Buffer
+		if err := png.Encode(&buf, dst); err != nil {
+			return nil, fmt.Errorf("annotategrid: encode: %w", err)
+		}
+		return buf.Bytes(), nil
+	}
+
+	w := bounds.Dx()
+	h := bounds.Dy()
+	lineColor := color.RGBA{R: 220, G: 30, B: 30, A: 160}
+
+	// Draw vertical grid lines.
+	for c := 1; c < cols; c++ {
+		x := c * w / cols
+		for y := bounds.Min.Y; y < bounds.Max.Y; y++ {
+			dst.SetRGBA(x, y, lineColor)
+			if x+1 < bounds.Max.X {
+				dst.SetRGBA(x+1, y, lineColor)
+			}
+		}
+	}
+
+	// Draw horizontal grid lines.
+	for r := 1; r < rows; r++ {
+		y := r * h / rows
+		for x := bounds.Min.X; x < bounds.Max.X; x++ {
+			dst.SetRGBA(x, y, lineColor)
+			if y+1 < bounds.Max.Y {
+				dst.SetRGBA(x+1, y+1, lineColor)
+			}
+		}
+	}
+
+	// Draw cell labels.
+	face := basicfont.Face7x13
+	ascent := face.Ascent   // plain int pixels
+	descent := face.Descent // plain int pixels
+	advance := face.Advance // plain int pixels per glyph
+	labelColor := color.RGBA{R: 255, G: 255, B: 255, A: 255}
+	bgColor := color.RGBA{R: 0, G: 0, B: 0, A: 180}
+
+	for r := 0; r < rows; r++ {
+		for c := 0; c < cols; c++ {
+			idx := r*cols + c
+			label := fmt.Sprintf("%d", idx)
+
+			cellX := c * w / cols
+			cellY := r * h / rows
+
+			// Label position: 2px padding from cell top-left.
+			px := cellX + 2
+			py := cellY + ascent + 2
+
+			labelW := len(label) * advance
+
+			// Draw dark background rect behind label.
+			bgRect := image.Rect(px-1, py-ascent-1, px+labelW+1, py+descent+1)
+			draw.Draw(dst, bgRect, &image.Uniform{bgColor}, image.Point{}, draw.Over)
+
+			// Draw label text.
+			d := &font.Drawer{
+				Dst:  dst,
+				Src:  image.NewUniform(labelColor),
+				Face: face,
+				Dot:  fixed.P(px, py),
+			}
+			d.DrawString(label)
+		}
+	}
+
+	var buf bytes.Buffer
+	if err := png.Encode(&buf, dst); err != nil {
+		return nil, fmt.Errorf("annotategrid: encode: %w", err)
+	}
+	return buf.Bytes(), nil
+}
+